• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

SOLVED FX 8350 - Curious Overheating Issue

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

BoundByBlood

Maybe Something Cool?
Joined
Sep 27, 2009
Location
MS Gulf Coast
I have become convinced that one of my most recent issues was caused by my cpu overheating when under heavy workload such as gaming. I don't have proof of this since I wasn't able to record temps, but it does make sense considering the heatsink doesn't have heatpipes and the contact area on the base of the heatsink didn't fully cover the IHS plate on the processor [I have since removed said heatsink].

I was thinking about something which struck me as a little odd and here is where it gets strange. With the crappy heatsink installed I did run all the tests in prime95 for about an hour each and it didn't fail. So I was wondering how could the chip overheat under load in gaming, but not do so when under maximum stress in prime95? Isn't that a bit of an impossibility?

Or am I off base here and can such a thing happen? Has something similar to this happen to anyone?
 
Last edited:
I would guess that your problem is either with the graphics card and/or the PSU no longer providing enough voltage under load. Changing the CPU HSF was a good idea. Those FX chips can produce a lot of heat. Prime95 will only test CPU and/or memory under load but not the GPU. Something like 3D Mark will stress both.

When decieding which part to suspect first, look at the age of each part. If you have had that HDD for a while, maybe it is beginning to fail. Test the RAM with MEMTEST86+. Make sure fans are working and that cooling fins are clean on all parts like the CPU, GPU and MB.
 
Actually this was one of the most complex and confusing issues and I do have a thread about this specific issue in the AMD GPU subtopic. I have already sent the GPU off on a RMA.

look at the age of each part
Everything is brand new retail in the box.

Something like 3D Mark will stress both.
The R9 390X would pass 3DMark 11 with flying colors, but the PC would still lock with the buzzing sound looping through the speakers during gaming.

Test the RAM with MEMTEST86+
I tested all the components - prime95 was without error as was memtest, I tested the HDDs with HDTune and WD's Data LifeGuard - everything passed. The only thing I could not physically test myself was the PSU, also new, because I just didn't have the tools, but it was working fine and as far as I could tell the PSU didn't fail such as when stress testing the CPU/RAM in Prime95 or the GPU in 3DMark. This tells me it was able to provide voltage to components under load.

I created this thread to see if it was possible a CPU could overheat and fail under load in game, but not during prime95.
 
Your original post gives no info for us to help. A little more info would be nice ...... ie:

What type of cooler? .... Stock?
Overclocked or Stock settings?
CPU Voltage?
Other voltage settings?
Number of fans in your case and the type of fans?
 
Your original post gives no info for us to help. A little more info would be nice ...... ie:

What type of cooler? .... Stock?
Overclocked or Stock settings?
CPU Voltage?
Other voltage settings?
Number of fans in your case and the type of fans?

[1] I'm putting a wraith on it.
[2] Stock
[3] Default
[4] Default
[5] 3x120mm front intake, 2x120mm top exhaust [cooler master], 1x140mm rear exhaust. The front intake and rear exhaust came with the case, Rosewill Rise, and I added the top exhaust. The cooler master fans are about 60-65 CFM, not sure about the others.
 
Gaming requires a lot of resources. HDD read and writes, RAM read and writes. CPU usage pretty high depending on game. Larger L3 cache loads (creates good heat too) Video card. So the motherboard gets warmer using the tracings in the PCB from the NB SB to the PCIE lanes and so on.

P95 is good for a calculation for just the CPU and a load on memory. Not the tell all solution be any means.
 
At least in the past, I found Linpack to give higher CPU temps than Prime95.

Linpack seems to be the best for core testing and Prime95 best for system bus and RAM.

Another test is the x.264 HD 5.0.1 benchmark. If you have heat or PSU issues, it should shut down during the x.264 test.
 
I am not sure your chip overheated while gaming. Just my thoughts.

When your GPU comes back give it another go open HWMontor in the background and give it a go for 30 min to an hour and then check your temps. If they are high then maybe I am wrong, and post back with some results. Something you can try if you feel like it, is add a couple fans inside your case 1 blowing on the VRM / NB cooler to the left of your CPU as well as adding one in behind your CPU Socket area if you have the ability to do so. These two fans will help with CPU Socket temps / Motherboard temps.
 
I know this much that is after upgrading the HSF on my cpu and when I get my GPU back from Asus, if it still freezes then it's undoubtedly the PSU. It literally cannot be anything else at this point.
 
I know this much that is after upgrading the HSF on my cpu and when I get my GPU back from Asus, if it still freezes then it's undoubtedly the PSU. It literally cannot be anything else at this point.

See we really need to know what the deal is here.

SO let's straighten it out just to be 100% sure what the problem is then we can go from there.....

BSOD = Often times a memory and bus frequency "crash"

Freeze up = screen hang no response. This can be often times as simple as a low cpu v-core hang up. (major problem with FX-9590) A faulty HDD can cause this as well.

PC shuts off = over heat protection via 85c core temp. This shuts the PC off on most FX processors. This is CPU temp (cores/package) not the socket temp.

These are the most common of problems.

Which one of these are happening? = If CPU is being manually driven even at stock, it would be suggested by myself to run one click v-core higher than P-state recommendation for X clock speed.

In example:

#1: 4200 MHz, 1.425V
#2: 4100 MHz, 1.4V

If we where to pick say 4200mhz P-state, this one only loads 4 cores on light to medium demands. But while running all 8 cores, you'd want to take this v-core up a click perhaps as high as 1.4500v to make every core stable at this speed.

Cool and Quiet is usually recommended to be turned off. With a FX-9590 for example, it creates problems because at the nano second level, the CPU doesn't receive enough voltage and it creates a screen freeze or hang up.

I'm not super familiar with Gigabyte boards even though I'm typing on one right now. The bios is somewhat buggy compared to an Asus board. It seems to take a little more work to get a stable system for some reason. However I'm not running an FX processor at this time, but found a little trouble with OC Opteron 1389 to 3.4ghz and higher because of the voltage warning BS the board has implemented. Always flashing that voltage is not properly adjusted for the overclock or some crazy stuff. On an Asus board, this doesn't exsist. If the OC fails, it simply reboots, tells you about it and you make adjustments. But with the GIgabyte board, it seems you have to please the bios to get it running nice. It's weird to say the least.

Now since you're running stock, we'll have to wait and see when you get the Wraith cooler and video card back from RMA and let you toy with it for a bit. Then try and find a solution if there's a problem still reoccurring. If a freeze still exists, perhaps a test to the HDD wouldn't hurt any?

Goodnight and good luck!

Shrimpinabed~ :p
 
How long of a Memtest did you run on the RAM sticks? I have your motherboard (sig outdated) and two of my slots are straight up dead, so there may just be a RAM problem in all of this. If the game just hangs and the GPU has passed tests, the CPU has passed tests, but you haven't stressed the mem only, memtest is a nice way to do that. Can you also turn down the MEM freq in the BIOS and make sure its running at rated voltage using CPU-Z or at least, again, through the BIOS?

I have found this feature with AIDA64 that will stress all the components at once and it used to crash my system when I ran it. It loads up the MEM to about 90%, runs the CPU %100 and GPU %100, but I don't use it for thermal testing.

I will say that my 9370 which is just an oc'd 8320 has to have a backplate fan which I modded into my case and is based off of the CPU temps otherwise my rig gets in trouble with temps really quick.

One quick question, as if the PSU has "failed" you shouldn't get anything back from your computer. Have you plugged the computer directly to the wall or a UPS unit? The quality of the input signal can impact the signal generated by clock sources on the mobo.
 
BSOD = bus frequency "crash"

Especially with Prime95 in blend mode...



PC shuts off = over heat protection via 85c core temp. This shuts the PC off on most FX processors. This is CPU temp (cores/package) not the socket temp.

For me, it looks like mine may be shutting down because of reaching 70 C on the "socket" temp. Is this a common Sabertooth 990FX R 2.0 issue???
 
Especially with Prime95 in blend mode...





For me, it looks like mine may be shutting down because of reaching 70 C on the "socket" temp. Is this a common Sabertooth 990FX R 2.0 issue???

Probably not. Thermtrip is processor protection. However, the motherboard may shut down to protect VRM from various things. Possibly surge protection which usually bios notifies, or maybe VRM or NB over heat protection. Also a Psu shut down can occur. The way to tell the difference is when the PSU shuts down, you'll have absolutely no power to the motherboard at all and requires a shut down on the switch of the PSU sometimes taking a minute for a reset the power the PSU back on.
 
Which one of these are happening?
Freeze Up would be the most accurate description. The PC will freeze in the middle of the game - image stays locked on screen - and there is a buzzing sound looping through the speakers. At the bottom of this post will be a video of the exact issue I'm having I found on youtube. Even the noise is the same.
Cool and Quiet is usually recommended to be turned off.
When I get everything back I'm turning off all advanced power savings features in BIOS. This board has a ton of them - C1E, C6, CnQ, APM - and I'm going to try running the machine without them.
On an Asus board, this doesn't exsist. If the OC fails, it simply reboots, tells you about it and you make adjustments. But with the GIgabyte board, it seems you have to please the bios to get it running nice. It's weird to say the least.
This board will actually reboot and warn you that it didn't pass POST and then gives you the option to go back into the bios.
perhaps a test to the HDD wouldn't hurt any?
I have tested my drive with WD's Data LifeGuard and HDTune...no problems. I don't really have any reason or exigent circumstance to suspect the drive at this point.
How long of a Memtest did you run on the RAM sticks?
About an hour.
two of my slots are straight up dead, so there may just be a RAM problem in all of this
Could be, but unlikely. When I go into bios it recognizes all four of the modules and they are all running in dual channel. Usually when there is a bad ram slot the bios will show that slot as not being populated [one of the many tell factors that my old MB was going bad], I don't have that problem.
Can you also turn down the MEM freq in the BIOS and make sure its running at rated voltage using CPU-Z or at least, again, through the BIOS?
When I put this system all together I put 1866 ram in there, but this board only supports 1866 in two modules so right off the bat it underclocked the ram to 1600 using the default voltage for those timings.
One quick question, as if the PSU has "failed" you shouldn't get anything back from your computer. Have you plugged the computer directly to the wall or a UPS unit?
Once I get my card back and I put it all back together I'm going to plug the PSU straight into the wall instead of the surge protector.

Here is the video with the exact issue I'm experiencing even with the same noise. This guy says it only happens to him in DOTA, I experience this in all modern D3D games [browser games are an exception]:
 
Last edited:
Probably not. Thermtrip is processor protection. However, the motherboard may shut down to protect VRM from various things. Possibly surge protection which usually bios notifies, or maybe VRM or NB over heat protection. Also a Psu shut down can occur. The way to tell the difference is when the PSU shuts down, you'll have absolutely no power to the motherboard at all and requires a shut down on the switch of the PSU sometimes taking a minute for a reset the power the PSU back on.

Nothing indicated a toasty VRM and still a shutdown... The VRM looks very cool! I suspect it's Asus hardcoding a low shutdown temp...
 
The both of you have interesting problems. To tackle BlackHeat's, we need his system back up and running. From previous posts and threads, we are banking on a video card problem, but addressing any cooling problems in the meantime is a good idea any ways.

_________________________________________________________________________________________________

RJARR..... the way you put hardcoding and low shut down temp is interesting. I'd relate this to a report rather than say it's hard coded.

All I can tell you from any experience I've had with probably a good 10 or 12 FX processors, I've never seen what you'd call a low temp shut down. I'm so familiar with Asus over any other board, it's not funny. I could write a very long winded post about Asus motherboards and their functions in great detail, but it won't solve issues in most cases.

If I where you, I'd test the PSU. Firstm a multimeter on the 12v rails. Then test the 5v rails. You get a dip exceeding more than say 5-10%, that PSU is no good. Another way to test is by trying a spare PSU you might have laying around that is in good standing.

Speaking of power supply, multiple rail PSU's need to be balanced so to speak. Let's say you have 3 12v rails instead of one. If you load One of those three rails too much more than the other two, it may cause issues like random shut downs. I've seen this many times.

A PSU overheat would cause a shut down. I've actually upgraded the fan on my 1000w Antec Continuous Power Series PSU because it would run hot while using LN2 and pushing 2.0v on the CPU. The wires would actually even get warm. The upgrade also helped with running high wattage TECs (which is one of my things I like to do) and that also puts a very heavy load on the PSU. And I mean big time. You could try and reconfigure the peripherals attached to your PSU and see if that helps any.

After reviewing your SIG really quick, this is what I see.

+3.3V@30A; +5V@28A; +12V1@18A; +12V2@18A; [email protected]; [email protected] 500W

500W = max power output. Efficiency > 85% = 85% typical load capacity. This PSU may just not be big enough.

GTX 660ti =
Maximum GPU Tempurature (in C) 97 C
Maximum Graphics Card Power (W) 150 W
Minimum System Power Requirement (W) 450 W

The FX-8350 uses roughly 210W at load 4ghz. At 4.8ghz around 385W

The 660ti (stock load) and FX-8350 (4ghz load) together at load are using 360W just for these two pieces of hardware. We are not counting the HDDs, motherboard DVD rom, fans and so on.

85% of 500 is 425W. you'd think that 500W is plenty, but in this case, I don't think it is. You have 65W of head room if you where to load CPU and Video card.

So after crunching some numbers there, You may want to slim the power usage (even though we haven't gone into amp usage yet) and see if you can cut some power here and there, or go with a larger power supply.

Running a FX processor with 8 cores is no joke. Always look for over kill with 2 things concerning a FX processor. 1 cooling and 2 PSU.

650w - 850w closer to 850W would be ideal. Leaves room for extra HW, SLI or crossfire and overclocking.

How important is a PSU for overclocking? Well VERY important. Lemme give an example. While running LN2, I gained 200mhz upgrading the psu from 850w to 1000w during a max clock LN2 run. That would be a jump from 7450mhz to 7685mhz with the exact same set up nothing changed except the PSU.

Long winded post. But I hope you take the time to read it all. ^^ :chair:
 
Last edited:
A failing PSU may still provide some power but not enough to carry the system under load or even to finish booting. Doesn't necessarily mean it's completely dead.
 
The shutdown incident was when I had a GT 640 in the rig... I'm currently at stock...

The FX 8 cores are monsters.

I'd even still recommend a larger PSU even with the lowest end of video cards.

Haven't run anything under 850w in over 6 years. Since Phenom II started eating up my 500w and 650w PSU's, I said F'it and go big or stay home.

Since upgrading to much larger PSU of sorts, I've never had a power issue.

Save some cash and shops around brother. Trust me, a nice big good power supply will go a very long way. You'll even be able to overclock (given you have some decent cooling there) and won't have any issues. If you get another shut down, you'll know for sure it's not that nice shiny new PSU... unless you get unlucky and get a dud.

Seasonic 80+ bronze 105$ 850W
Efficiency 82% Energy-Efficient
80 PLUS BRONZE Certified
http://www.newegg.com/Product/Product.aspx?Item=N82E16817151108

Note the AMPs on that 12v line. You'll go from 18 amp per 2 12v lines you have now to 70A on one line. Run the entire rig on a single 12v rail.

I like Antec. My next one may be... Antec 1300W 289$
Efficiency 94% (MAX) Energy-Efficient
80 PLUS PLATINUM Certified
http://www.newegg.com/Product/Product.aspx?Item=9SIA25V2SC4692

The difference between these two is very great. I like Seasonic though. Very stable and reliable power supplies. If I was building a daily rig right now, I'd go with the one there I linked ya. Way overkill. Will never run hot unless you get into SLI and some big wattage GPUs and overclocking that CPU with it. Other than that, it'll serve you many years.
 
Well this is simple. Its the motherboard.

benching pulls the most power out possible, so passing that implies something else is going on. Sudden blackouts were the precursors to all my motherboards failing with my 8320. Could pass benchmarks all day, but random crashes (complete black out) lead to me looking at the motherboard.

So thats my bet, but a better thing to do is look up your windows logs. Its been a while since i've done it, someone else can chime in on that.

edit: could be the ram as well. FX chips are finnicky with multiple sticks of ram.

edit 2: enermax makes decent psu's, far cry from the diablotek crap. 1000w is way overkill for his uses as well. I ran my 8320 @5.0 ghz and an overclocked 290 with no issues at all on a 620w psu.

edit 3: yeah looking at that failure video, thats not a power supply issue. PSU will simply black out the screen and restart the computer. Your windows logs should definitely help narrow down this issue.
 
Last edited:
Back